<p>This report presents Frigate, a novel astronomical dataset comprising still-frame images of a section of Low Earth Orbit (LEO), collected by ExoAnalytic Solutions Inc., a commercial space situational awareness (SSA) provider using ground-based optical systems. Datasets of this kind are rare and typically inaccessible to most stakeholders. Through agreement with the company, the images were collected and prepared for use in machine learning and computer vision tasks related to SSA. The raw images contain lens glare artefacts, vignetting, and outer-region degradation. To address this, we additionally release a lightly processed version of each image featuring artefact removal, background subtraction, and sharpening. Usage notes describe optional downstream techniques including stacking, streak detection, and automated annotation. Given the lack of publicly available space surveillance data, this dataset represents a valuable resource for satellite tracking, orbital debris detection, and astronomical computer vision. The dataset described in this paper is wholly owned by ExoAnalytic Solutions Inc. and was released to the research team through mutual agreement of the benefits of its processing and application to the domain. All data and code are openly available.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Frigate: A Novel Dataset Comprising Wide Field-of-View Astronomical FITS Images of Low Earth Orbit Scenes for Machine Learning Applications

  • Daniel S. Roll,
  • Zeyneb Kurt,
  • Yulei Li,
  • Ralph Dinsley,
  • Wai Lok Woo

摘要

This report presents Frigate, a novel astronomical dataset comprising still-frame images of a section of Low Earth Orbit (LEO), collected by ExoAnalytic Solutions Inc., a commercial space situational awareness (SSA) provider using ground-based optical systems. Datasets of this kind are rare and typically inaccessible to most stakeholders. Through agreement with the company, the images were collected and prepared for use in machine learning and computer vision tasks related to SSA. The raw images contain lens glare artefacts, vignetting, and outer-region degradation. To address this, we additionally release a lightly processed version of each image featuring artefact removal, background subtraction, and sharpening. Usage notes describe optional downstream techniques including stacking, streak detection, and automated annotation. Given the lack of publicly available space surveillance data, this dataset represents a valuable resource for satellite tracking, orbital debris detection, and astronomical computer vision. The dataset described in this paper is wholly owned by ExoAnalytic Solutions Inc. and was released to the research team through mutual agreement of the benefits of its processing and application to the domain. All data and code are openly available.